Post on 23-Jul-2020
The Smart Timing Choice™ 1 SiT-AN10045 Rev 2.2
March 2015
Resilience and Reliability of Silicon MEMS Oscillators
Table of Contents
1 Introduction ......................................................................................................................................... 2
2 Silicon MEMS Advantages ................................................................................................................. 2
3 Environmental Stressors .................................................................................................................... 3 3.1 Power Supply Noise ................................................................................................................... 4 3.2 External EMI Noise .................................................................................................................... 5 3.3 Shock and Vibration ................................................................................................................... 8
4 Reliability .......................................................................................................................................... 10
5 Summary .......................................................................................................................................... 11
6 References ....................................................................................................................................... 12
The Smart Timing Choice™ 2 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
1 Introduction
Oscillators have historically been made from quartz crystal resonators connected to an analog
sustaining circuit that drives the resonator to vibrate at a specific frequency. Now, there is an
alternative – Silicon MEMS oscillators – and these devices outperform quartz oscillators in
noisy environments. The drive toward higher speed telecom and mobile applications places
greater demands on the clock source. Additionally, more complex electronics and higher clock
frequencies necessitate that the clock device continue to perform well in noisy environments.
This paper shows results of comparative experiments that were conducted on quartz and
Silicon MEMS oscillators. The data demonstrate that MEMS oscillators outperform quartz in
realistic environmental conditions.
Oscillator vendors provide data sheets for each product stating performance parameters such
as frequency stability, jitter, and phase noise. While data sheets are a good indicator for
selection of timing devices, the user must also evaluate how these devices perform in real-life
environmental conditions. Testing under conditions that mimic those seen in the real operating
environment provides valuable information about true component performance. The
performance of oscillators subjected to environmental stressors, such as electromagnetic
interference (EMI), vibration, and noise from power supplies or other system components, will
degrade as compared to oscillators in ideal conditions. Ultimately, environmental stressors
may reduce the reliability and lifetime of a device. It is important to consider the performance
of oscillators under realistic, noisy, harsh conditions when selecting a timing device.
2 Silicon MEMS Advantages
Silicon MEMS oscillators have some inherent advantages over quartz oscillators that allow them
to perform reliably in a variety of environments. SiTime developed the MEMSFirstTM process, in
which resonators are fully encapsulated in silicon and enclosed within a micro-vacuum chamber
[1]. The combination of very small mass of the resonator and its stiff silicon crystal structure
makes them durable and extremely resistant to external stresses such as shock and vibration.
Additionally, optimally designed analog circuits in the oscillator deliver high performance in
electrically noisy conditions.
The schematic of MEMS oscillator architecture in figure 1 shows the key components that
contribute to performance and reliability, including a finely tuned silicon MEMS resonator,
oscillator sustaining circuit, high precision fractional-N phase locked loop (PLL), and drivers
with fully differential circuits.
The Smart Timing Choice™ 3 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
OE / STBY / PROG
PLL
Vbias
GND
Oscillator Die
VDD
GND
CLKFractional-N
PLL
Output Dividers
And Drivers
Charge
Pump
Vbias
Resonator
Sustaining
Circuit
Digital
Temperature
CompensationA/DTemperature
Sensor
I/OOTP Memory
MEMS
Resonator
Figure 1. SiTime MEMS oscillator architecture
Most quartz oscillator vendors are experts in manufacturing resonators, but not necessarily in
circuit design. They usually outsource the analog circuit and must purchase die that are
designed to work with a variety of crystals rather than being optimized for the specific
resonator. In contrast, SiTime has a world-class team of analog designers who design all the
circuits that are used with SiTime’s MEMS oscillators. Since 2006, this team has made
dramatic improvements in performance and resilience of SiTime’s oscillator products with the
result that SiTime’s MEMS oscillators are more resilient than quartz devices in noisy
environmental conditions.
3 Environmental Stressors
Several factors in the operating environment can negatively impact oscillator performance,
degrading phase noise and jitter. Taking each in turn, this paper will compare the effect of
environmental conditions on the performance of oscillators produced by SiTime and
competing manufacturers.
The Smart Timing Choice™ 4 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
3.1 Power Supply Noise
One major source of noise in any system comes from power supplies. Most of this noise is
filtered out by passive filters and decoupling capacitors that are placed on the power supply
input of the oscillator. However, some noise remains, which may increase the jitter on the
output clock, and can negatively impact system timing margins. This noise is amplified not
only when the power supply itself is switched on, but when other devices on the board turn on
or off during system operation. On-board issues, such as inadequate power supply filtering or
ground bounce, also affect noise and jitter. The Power Supply Rejection Ratio (PSRR) is a
specific parameter that is used in the design of analog circuits and provides an indication of
how robust a circuit is to noise from the power supply. Unlike PSRR which is a SNR related
parameter expressed in dB, oscillator performance under noisy power supply conditions is
expressed by the Power Supply Noise Sensitivity (PSNS) metric. PSNS is quantified in terms
of the induced phase jitter exhibited by the oscillator when subjected to a controlled peak-to-
peak noise injection at specific frequencies in the range of 20 kHz to 20 MHz.
1 nF 10 kΩ
1 uFVDD
OUT-
GND
OE
10 Ω
50 Ω
1 μF
Power
Supply
220 nH
Noise Source (Agilent 33250A)
Oscillator under
test 12GHz active
probe
To
oscilloscope
Fast edge
comparator
To Agilent
DSA90604
oscilloscope
OUT+
AC/DC combiner for
adding supply noise
Figure 2. Block diagram for power supply noise rejection test set-up
A test setup that includes a power supply and waveform generator, as shown in figure 2, is a
controlled test method to evaluate the PSNS performance of oscillators. The waveform
generator adds system noise at a specified voltage and frequency to measure the effect of
power supply noise on the oscillator jitter. The plot in Figure 3 shows integrated phase jitter as
a function of power supply switching noise frequency for 50 mV of peak-peak power supply
The Smart Timing Choice™ 5 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
noise, comparing results across various quartz oscillators with those of a SiTime MEMS
oscillator for LVCMOS outputs. As the plots indicate, SiTime's MEMS oscillator jitter is lower
across all noise frequencies. The reason for this is noise-reducing circuits built into SiTime’s
oscillator circuitry to protect the oscillator from power supply-induced jitter.
Figure 3. Phase jitter in the presence of 50 mV power supply noise for SiTime MEMS and
Epson SAW oscillators as a function of power supply switching noise frequency
3.2 External EMI Noise
Another important noise source to consider is externally generated EMI noise that impacts the
oscillator performance (as opposed to EMI signals that are emitted by a clock source). Power
supplies, power lines, lightning, computer equipment, and electronic components are all
potential sources of externally generated EMI, which can be coupled into the system through
radiation. EMI is a major concern in applications such as passive optical networks (PON),
cellular base stations and many products that are used in outdoor environments where large
electromagnetic sources are present. EMI is also a concern in dense electronic boards with
multiple switching power supplies, since oscillator components can be placed close to these
power supplies. Inbound EMI can change the clock jitter and, in catastrophic cases, even the
operating frequency of clock devices, negatively impacting the functioning of any system that
depends on the clock signal for reliable performance. Phase jitter and phase noise increase
significantly in the presence of incoming EMI, and attempts to filter out the noise reaching the
0
1
2
3
4
5
10 100 1,000 10,000
Inte
gra
ted
Ph
ase J
itte
r p
er
mV
p-p
in
jecte
d n
ois
e (
ps/m
v)
Power Supply Noise Frequency (kHz)
Power Supply Noise Rejection
SiTIme NDK Epson Kyocera
SiTimeUp To 7x
Better
The Smart Timing Choice™ 6 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
oscillator are not always successful. Another approach is to design clock devices that
successfully reject EMI. Electromagnetic susceptibility, or EMS, quantifies the detrimental
impact of EMI on electronic circuits such as oscillators.
EMS can be measured by following the procedure specified in EMC standard IEC EN61000-
4.3. This standard specifies a radiated electromagnetic (EM) field of 3V/m across a frequency
range of 80 MHz to 1 GHz in 1% incremental steps. The device under test is located in a
calibrated anechoic chamber and positioned so that it is aligned with the axis of the vertically
polarized antenna, as shown in figure 4. The phase noise analyzer and high precision, low
noise digital signal analyzer capture the oscillator phase noise. The electromagnetic field
induces noise spurs, and the average power of the spurs provides a measure of the EMS of
the oscillator.
Figure 4. Setup for EMS Testing
The Smart Timing Choice™ 7 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
Data on multiple quartz and SiTime MEMS oscillators illustrate the effect of EMI on both
differential and single-ended oscillators (Figures 5 and 6). SiTime MEMS oscillators
outperform competing quartz- and MEMS-based oscillators by a considerable margin. These
results emphasize the importance of understanding the relationship between performance and
operating environment.
Figure 5. Average level of EMI-induced phase noise spurs on 156.25MHz LVPECL
differential clock oscillators
Figure 6. Average level of EMI-induced phase noise spurs on 26 MHz single-ended oscillators
For details on oscillator EMS performance test conditions and experiment results, see “Electromagnetic Susceptibility Comparison of MEMS and Quartz-based Oscillators” [2].
The Smart Timing Choice™ 8 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
3.3 Shock and Vibration
Many electronic products are subjected to substantial vibrational forces during use. This is
especially true of mobile, portable devices carried around in pockets or backpacks. Electronics
in mobile GPS units, industrial equipment, or aerospace applications may undergo higher levels
of vibration. Even stationary products may experience vibration from a nearby fan or other
equipment.
Quartz oscillators may show significant sensitivity to vibration because of the mechanical
assembly and packaging used. SiTime's MEMS FirstTM technology [1] produces MEMS
resonators that are inherently more resistant to vibration-induced performance degradation for
two reasons. First, the moving section of the silicon resonator has much smaller mass than a
quartz resonator. This reduces the force applied to the resonator from the vibration-induced
acceleration. Second, the silicon MEMS resonators are very stiff structures that vibrate in-plane
and are therefore resistant to movement caused by vibration forces.
Vibration can degrade oscillator performance by inducing an electrical signature at the same
frequency as the mechanical vibration, causing frequency spikes or increased phase jitter or
broadband noise. Mechanical forces may also damage the physical resonator structure.
Because oscillator response depends on the direction, severity and frequency of external
mechanical forces, looking at results from several different types of tests gives the most
complete picture of the resiliency of an oscillator.
The first test is evaluating the response to sinusoidal vibration by observing spurious phase
noise, or noise spurs, occurring at a specific frequency. This phase noise is translated into a
frequency modulated (FM) noise and normalized to the carrier frequency for 1g vibration
acceleration. The result is expressed in part-per-billion/g (ppb/g) as a function of vibration
frequency. The measurement setup includes a controller, power amplifier and shakers.
Sinusoidal vibration tests included vibration frequencies ranging from 15 Hz to 2 kHz with a
peak acceleration of 4g, representative of vibration forces that oscillators would experience in
the field. Oscillators were subjected to vibration in the x, y and z axes, and results reported are
the highest observed noise response for the three directions.
Figure 7 presents vibration sensitivity results for quartz-, SAW- and MEMS-based differential
oscillators. SiTime MEMS oscillator outperformed other devices by a factor of 10 to 100.
The Smart Timing Choice™ 9 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
Figure 7. Differential XO maximum vibration sensitivity vs. frequency under 4g peak
acceleration sinusoidal vibration in X, Y, or Z axis
The second vibration test generates random vibration at an acceleration of 7.5-g rms, as
defined by MIL-STD 883F, which simulates a harsh operating environment. Results are
reported in terms of induced phase jitter in the time domain, calculated by integrating the
induced phase noise over 15 Hz to 10 kHz offset frequency. The data (see figure 8) show a
wide range of responses. The SiTime MEMS oscillator outperforms all the other devices,
showing that it is relatively immune to random vibration.
Figure 8. Differential oscillator phase jitter induced by random vibration
SiTime
Up to 18x Better
The Smart Timing Choice™ 10 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
SiTime MEMS oscillators are also insensitive to shock impact, the third test of resistance to
mechanical force. Sudden shock tends to cause transient deviations in oscillator frequency.
SiTime measured the response of quartz- and MEMS-based oscillators to a 1 ms half sine
wave shock pulse with an acceleration of 500 g. Results in figure 9 show that while most
devices exhibit significant frequency deviation, the frequency of the SiTime MEMS deviates by
less than 1 ppm.
For details on oscillator shock and vibration performance comparison test conditions and
experiment results, see “Shock and Vibration Performance Comparison of MEMS and Quartz-
based Oscillators” [3].
Figure 9. Response of differential oscillators to 500-g shock
4 Reliability
One method of quantifying the reliability of a component is predicting mean time between
failure (MTBF). For semiconductor components, this is the inverse of the failures in time (FIT)
rate, expressed as the number of failures statistically expected after 1 billion operating hours.
The higher the MTBF, the longer the expected lifetime of the device and therefore the more
reliable the device. Low values of FIT rate indicate a low number of expected failures and high
reliability.
SiTime calculates FIT by subjecting oscillators to stress testing at elevated temperature and
voltage for an extended period of time. Stressing thousands of oscillators for cumulative test
The Smart Timing Choice™ 11 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
time of over two million device hours resulted in no failures. Based on these results, SiTime's
MEMS oscillators have a calculated reliability rate of less than 1 FIT, corresponding to a
MTBF of 1000 million hours. For details on methods to calculate FIT rate and MTBF, see
"Reliability Calculations for SiTime Oscillators" [4].
Figure 10 includes reported MTBF for oscillators from various manufacturers, showing that
SiTime’s MEMS-based oscillators are significantly more reliable that quartz oscillators.
Figure 10. Reliability of quartz- and semiconductor-based oscillators in terms of MTBF
5 Summary
It is important to consider oscillator performance in real-world conditions in order to truly
understand device capabilities. Actual operating conditions often include the presence of
power supply noise, external EMI, vibration and shock. Testing under these conditions
demonstrates the advantages of SiTime MEMS oscillators as compared to competing quartz
and semiconductor-based devices. SiTime outperforms the competition in all four categories
considered.
Power supply noise: up to 7x better than quartz
Susceptibility to EMI noise: up to 50x better than competitors
Vibration: up to 40x better than quartz
Reliability: up to 80x better than quartz
14
16
28
38
1000
0 200 400 600 800 1000 1200
Pericom
TXC
Epson
IDT (Fox)
SiTime
Mean Time Between Failure (Million Hours)
1000
The Smart Timing Choice™ 12 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
Environmental stresses such as external sources of noise and vibration can adversely affect
the resilience, performance and reliability of clock devices. The silicon MEMS oscillators
produced by SiTime are resilient in the face of environmental conditions ranging from typical
to extremely challenging, maintaining phase noise and jitter specifications and exhibiting long-
term reliability in a variety of operating environments.
6 References
[1] SiTime Corp., “SiTime’s MEMS First Process”, Application Note 20001, http://www.sitime.com/support2/documents/AN20001-MEMS-First-Process.pdf.
[2] SiTime Corp., “Electromagnetic Susceptibility Comparison of MEMS and Quartz-based
Oscillators”, Application Note 10031,
http://www.sitime.com/support2/documents/AN10031-EMS-Comparison-MEMS-and-
Quartz-Oscillators.pdf.
[3] SiTime Corp., “Shock and Vibration Performance Comparison of MEMS and Quartz-based
Oscillators”, Application Note 10032,
http://www.sitime.com/support2/documents/AN10032-Shock-Vibration-Comparison-
MEMS-and-Quartz-Oscillators.pdf.
[4] SiTime Corp., “Reliability Calculations for SiTime Oscillators”, Application Note 10025, http://www.sitime.com/support2/documents/AN10025-SiTime-Reliability-Calculations.pdf.
Revision History
Version Release Date Change Summary
1.0 02/12/12 Original document.
2.1 04/1/14 Document was re-structured.
2.2 03/11/15 Updated shock, vibration and EMS graphs. Replaced indirect reference to quartz and other MEMS vendors. Update MTBF chart.
The Smart Timing Choice™ 13 SiT-AN10045 Rev 2.2
Resilience and Reliability of Silicon MEMS Oscillators
SiTime Corporation
990 Almanor Avenue
Sunnyvale, CA 94085
USA
Phone: 408-328-4400
http://www.sitime.com
© SiTime Corporation, 2008-2015. The information contained herein is subject to change at any time without notice. SiTime assumes no responsibility or liability for any loss, damage or defect of a Product which is caused in whole or in part by (i) use of any circuitry other than circuitry embodied in a SiTime product, (ii) misuse or abuse including static discharge, neglect or accident, (iii) unauthorized modification or repairs which have been soldered or altered during assembly and are not capable of being tested by SiTime under its normal test conditions, or (iv) improper installation, storage, handling, warehousing or transportation, or (v) being subjected to unusual physical, thermal, or electrical stress. Disclaimer: SiTime makes no warranty of any kind, express or implied, with regard to this material, and specifically disclaims any and all express or implied warranties, either in fact or by operation of law, statutory or otherwise, including the implied warranties of merchantability and fitness for use or a particular purpose, and any implied warranty arising from course of dealing or usage of trade, as well as any common-law duties relating to accuracy or lack of negligence, with respect to this material, any SiTime product and any product documentation. Products sold by SiTime are not suitable or intended to be used in a life support application or component, to operate nuclear facilities, or in other mission critical applications where human life may be involved or at stake.